Smart Science

ISSN: (Print) 2308-0477 (Online) Journal homepage: https://www.tandfonline.com/loi/tsma20

Load Frequency Control of Multi-microgrid System

considering Renewable Energy Sources Using Grey
Wolf Optimization

Srinivasarathnam c, Chandrasekhar Yammani & Sydulu Maheswarapu

To cite this article: Srinivasarathnam c, Chandrasekhar Yammani & Sydulu Maheswarapu (2019):
Load Frequency Control of Multi-microgrid System considering Renewable Energy Sources Using
Grey Wolf Optimization, Smart Science, DOI: 10.1080/23080477.2019.1630057

To link to this article: https://doi.org/10.1080/23080477.2019.1630057

Published online: 07 Jul 2019.

Submit your article to this journal

Article views: 8

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tsma20
SMART SCIENCE
https://doi.org/10.1080/23080477.2019.1630057

ARTICLE

Load Frequency Control of Multi-microgrid System considering Renewable

Energy Sources Using Grey Wolf Optimization
Srinivasarathnam c , Chandrasekhar Yammani and Sydulu Maheswarapu
Department of Electrical Engineering, National Institute of Technology, Warangal, INDIA

ABSTRACT ARTICLE HISTORY

Grey wolf optimization (GWO) algorithm is used in this paper for optimal tuning of PID controller Received 28 November 2018
gains used in secondary frequency control of an autonomous microgrid system and the multi- Accepted 6 June 2019
microgrid system operates in isolation. Gains of PID controller and Integral time absolute error are KEYWORDS
considered as control variables and fitness function, respectively. Renewable Energy Sources (RES) Microgrid and multi-
which are the high degree of non-linear such as Wind Turbine Generators (WTG), Solar Photovoltaic microgrid systems;
generators (SPV) are included in the test systems. The Diesel engine generator (DEG) and Battery renewable energy sources;
energy storage system (BESS) has been considered for immediate load frequency control sources dynamic stability; PID
during perturbation in the system frequency. Various scenarios are considered in this paper to controller; secondary
demonstrate the superiority of the proposed controller. Scenarios include only a sudden change in frequency control; grey wolf
load, wind power and solar power integration, simultaneous incorporation of RES and diverse load, optimization
parametric uncertainty have been considered. Dynamic response of the system and the cropped
numerical results ascertain the signature of the proposed GWO controller in lowering the deviations
and settling time. Prior-art controllers existing in the literature have been used to compare and
validate the obtained results.

Abbreviations: RES: Renewable energy sources; KE: Gain of DEG; BESS: Battery Energy Storage
System; KBES: Gain of BESS; LFC: Load Frequency Control; TWTG: Time constant of WTG; DG:
Distributed Generation; KWTG: Gain of WTG; SPV: Solar Photovoltaic; A: Swept area; DEG: Diesel
Engine Generator; TBES: Time constant of BESS; ACE: Area Control Error; TE: Time constant of
DEG; FC: Fuel Cell; β: blade pitch angle; WTG: Wind Turbine Generator; CP: Performance
coefficient; ΔF1: Frequency deviation in microgrid-1; ΔPWT: Change in output wind power; ρ:
air density factor; λ: tip speed ratio; Vrated: nominal wind speed; Δψ: Change in solar radiation;
R1: Speed regulation constant of microgrid-1; VW: wind speed; R2: Speed regulation constant of
microgrid-2.; Vcut–in: cut-in wind speed; ΔF2: Frequency deviation in microgrid-2; Vcut–out: max-
imum cut-out wind speed; ΔPC: Control signal to governor; Psize: Population size; PID controller:
Proportional-Integral-Derivative controller; ΔPL: Change in Load; T12: Synchronizing coefficient
between microgrid-1 and microgrid-2; iter: Current iteration; GWO: Grey wolf optimization;
itermax: Maximum iterations; Kpmax: upper bound of proportional gain ; Kpmin: lower bound of

CONTACT Srinivasarathnam c c.rathnam@student.nitw.ac.in Department of Electrical Engineering, National Institute of Technology, Warangal
506004, INDIA
© 2019 The Author(s)
2 S. C. ET AL.

proportional gain; Kimax: upper bound of integral gain; Kimix: lower bound of integral gain ;
Kdmax: upper bound of derivative gain; Kdmin: lower bound of derivative gain; ΔPTie–line: incre-
mental change in tie line power exchange between microgrid-1 and microgrid-2; ΔPDG: Change
in DEG output power; ΔPBES: Change in BESS power; T1: Governor Time constant; ΔPPV: Change
in Solar power; T2: Transportation delay time constant; Prated: Rated wind power output; D: Load
damping constant; PWT: Power from wind Turbine; H: Inertia Constant

1. Introduction power grid power cannot be accessible considering

The reliable indicator of the development of any coun-
try in the world is the per capita consumption of
electrical energy. Thus, the consumption of electrical
energy also decides the overall GDP of any country.
However, a large percentage of electrical energy is
produced from conventional fossil fuels such as oil,
coal, natural gas. From the estimated data, it is clear
that these fossil fuels will not last for more than
another 200 years. Owing to the faster rate of depletion
of fossil fuel, the variation between total electric power
production and total load demand is increasing day-
by-day causing erratic power supply to consumers. Not
only this, numerous places all over the world yet to be
electrified, due to the scarcity of electrical power gen-
eration. In addition to this, due to the combustion of
fossil fuels, the release of harmful gases such as CO2,
NOx, SO2, causing serious environment problem
majorly acid rains and global warming [1].
Considering these challenges, electric power utilities
around the world are investigating renewable energy
sources (RES) based power generation technologies.
The major advantages of power production from RES
are eco-friendly, clean energy production technology,
sustainable sources of energy. Not only, power genera-
tion from RES relieved from the insecurity of energy
sources but also can be used as a most useful resource
for power generation for the places where central
geographical issues [2,3].
Due to the instigation of deregulation in the power
industry has made an opportunity to distributed gen-
eration (DG) which has evolved the notion of the
microgrid. A Microgrid is a low/medium voltage grid
consists of Distributed generation (DG) micro-sources
such as micro-turbines, DEG, BESS, WTG, SPV gen-
erators, FC which are knitted together along with small
loads through feeders. Figure 1 represents the sche-
matic view of the microgrid system. The microgrid in
general functions in two modes, i.e. grid-mode (to an
MV distribution system) and autonomous/islanded
mode [4]. The potential benefits of microgrids are
reliable, flexible, upgradable, economic operation,
dynamic islanding and efficient [5]. Considering these
advantages in addition to sophisticated metering in the
distribution system, advanced communication technol-
ogies, modern control strategies have reformed the
conventional structure of distribution system into
multi-microgrid system over the past decade [6].
Figure 2 depicts the simplified schematic of multi-
microgrid system.
Frequency controller restores the frequency to the
nominal value, whenever power generation and its
demand fluctuates in the microgrid system.
Traditionally frequency control has been performed
on the generation side. The primary frequency control
operates in the timescale of tens of seconds with a

Figure 1. Schematic view of Microgrid System.


Figure 2. Schematic view of Multi-Microgrid System.
decentralized governor mechanism and does not
restore the frequency to the nominal value. However,
the secondary frequency control restores the system
frequency to nominal value but in the timescale of up
to a minute or so with centralized governor mechanism
[7]. In view of this, secondary frequency control is
generally adopted for operating the system at nominal
frequency.
However, an autonomous microgrid may consist of
RES such as wind power, solar power, which supply
highly wiggling power to the MG due to irregular
wind speed, solar radiation aberration. Due to huge
power penetration by RES to fluctuating load, diesel
engine generator contributes less power (capacity of
DEG is less) to the load, which leads to low inertia of
the system. Severe and consequential large oscillation
can be observed due to the low inertia system [8].
Thus, frequency deviations and voltage fluctuations
are observed in the microgrid system which are of
serious concern [9,10]. Since the RES are nature
dependent, these cannot be used for frequency con-
trol. Diesel generator is used for demand response,
but it cannot handle the sudden change in power
demand of a microgrid because its response time is
more. Hence, Battery energy storage system is used
for quick balancing. In order to regulate the frequency
deviation of the microgrid for any change in supply or
load, a controller is needed to ensure that the set
points of the Microgrid are adapted to optimal
requirement [11]
Furthermore, the DG parameters (resistance, induc-
tance, and capacitance), as stated in [12], gets affected as
a result of aging and thermal effect, which results in
parametric variation. Thus, if the real microgrid para-
meters and the assumed model differs, the secondary
frequency controller should able to tackle the parametric
uncertainty and stabilize the system [13].
In view of above, there is a need of robust load
frequency controller (LFC) to mitigate the oscillation
and to ensure that the dynamic performance of the
autonomous microgrid and multi-microgrid systems
is within the satisfactory limits even with parametric
SMART SCIENCE 3
uncertainty and irregular power supply from RES.
Considering these facts, an efficient and appropriate
LFC technique is required for stable operation of
microgrid system.
Meta-heuristic techniques are grouped into three
main categories, i.e. (a) Evolutionary algorithms (EA)
(b) Swarm Intelligence (SI) based algorithms (c)
Physics-based algorithms [14]. Evolutionary algorithm
mimics nature’s evolutionary principles to search opti-
mal solution. Differential Evolution (DE), Genetic
Programming (GP), Evolutionary Programming (EP)
and Biogeography-Based Optimizer (BBO) are some of
the EA. Physical rules are the foundation for Physics-
based algorithms. A few of this category of algorithms
are Galaxy-based Search Algorithm (GbSA),
Gravitational Search Algorithm (GSA), Black Hole
(BH) algorithm, Big-Bang Big-Crunch (BBBC), Ray
Optimization (RO) algorithm. The third category of
meta-heuristics is Swarm Intelligence techniques. The
social behavior of swarms, herds, flocks from nature is
being mimicked by these algorithms. Particle Swarm
Optimization (PSO) is the most popular SI method
developed from birds flocking. Grey wolf algorithm,
Firefly algorithm, Bat algorithm, Krill herd algorithm,
Cuckoo search are some of the SI-based algorithms.
Traditional-approached and artificial intelligence-
based methods for LFC problem can be found in the
literature. Authors in [15–17] have proposed a conven-
tional PI controller-based technique, Mallesham and et
al. have proposed Ziegler-Nichols-based Controller for
parameters tuning of PID controller [18]. The author
in [19] has proposed hybrid system consist of WTG,
DEG, fuel cell and acqu-electrolyzer. The hydrogen
generated from acqu-electrolyzer is used as input to
fuel cell. PI controller is used for frequency stabiliza-
tion. H∞ controller for minimizing the oscillations is
proposed by Goya et al. [20], internal model-based LFC
has been presented by authors in [21]. Swarm intelli-
gence-based techniques have been exploited for tuning
of PID gains in the literature such as Moth-Flame
optimized fuzzy PID controller has been proposed in
[22]. Das in [23] used Genetic Algorithm, Kumar in
4 S. C. ET AL.

[24] used biogeography-based optimization technique.
Esha Gupta and Akash Saxena in [25] used Grey wolf
optimization for Automatic Generation Control of two
areas’ interconnected power system consisting of ther-
mal units only. In this article, the author has verified
the superiority of the algorithm with load change in
steps of 10%, 20% and 25% in different areas. However,
the author has not verified the robustness of the algo-
rithm for parametric uncertainty of the system and
with the incorporation of highly fluctuating RES in
the system.
It is noticed that Grey wolf optimization has been
attempted in various disciplines such as engineering,
networking, environmental modeling applications,
machine learning, image processing, medical and
bioinformatics due to its impressive nature of explora-
tion and exploitation ability in locating the optimal
solution [26]. In view of this, GWO algorithm has
been attempted in this paper for fine tuning the gains
of PID controller for dynamic stability of the consid-
ered test systems with step load changes, incorporating
renewable energy sources and parametric variation.
2. Modeling of Microgrid System
The block diagram of stand-alone single microgrid and
multi-microgrid systems are presented in Figures 3 and 4
respective. Both single and multi-microgrid systems con-
sist of a DEG, WTG, solar photovoltaic system and BESS.
The load demand is mainly being supplied by the RES.
The proposed systems are very reliable because
whenever RES sources fail to feed the desired load
demand owing to their intermittent nature, the diesel
generator will acts as a cushion to deliver the balance
load demand. BESS is used for backup supply for a
short time duration for dynamic stability of the system.
The excess energy generated by the RES will be used by
the Battery for its charging. Further, microgrid systems
can be expanded in case the load demand is enhanced.
PID controller output is connected to Diesel engine
generators and BESS in each individual microgrid to
mitigate the frequency deviation in both the microgrids
and also variation in tie-line power flow by varying the
active power support, so as to make zero steady-state
error in frequency response and tie-line power flow
and also for obtaining a quicker steady-state response.
2.1. Diesel Engine Generator
Figure 5 represents the block diagram of the first-order
transfer function model of DEG. The equilibrium
between power demand and its generation in an auton-
omous microgrid due to variation in solar power and
wind power is maintained by DEG with speed gover-
nor control action.
2.2. Wind Turbine Model
Wind turbine is used to exploit the kinetic energy from
the wind energy and it is converted as mechanical
energy. This mechanical energy is then transferred to
the rotor of the generator. The wind turbine consist of
turbine-generator shaft mechanism, which is used to
convert the rotor rotation into electrical energy [7].
Equation (1) represents the mechanical output of the
wind turbine and is defined as follows [27–29].
1
PWT ¼ ρACP ðλ; βÞVW
3
(1)
2
where λ is tip speed ratio, ρ is air density factor (Kg/cu.m),
VW is wind speed (m/s), CP is power coefficient, A is Swept
area(m2 Þ and β is blade pitch angle.

8 9
> 0; Vw < Vcutin >
>
> >
>
< 0; Vw > Vcutout =
Pwt ¼ Prated ; Vrated  Vw  Vcutout (2)
>
>  >
>
w þ 0:3314Vw þ
6 5 4
>
: 0:001312V w 0:04603V
else >
;
3:687Vw  51:1Vw þ 2:33Vw þ 366
3 2

8 9
> 0; Vw < Vcutin >
>
> >
>
< 0; Vw > Vcutout =
ΔPwt ¼ 0; Vrated  Vw  Vcutout (3)
> ½0:007872V 5  0:23015V 4 þ 1:3256V 3 þ
> >
>
>
: w w w else >
;
11:061Vw2  102:2Vw þ 2:33:ΔVw
 SMART SCIENCE 5

Figure 3. Block diagram of stand-alone microgrid.

Figure 4. Block diagram of Multi-Microgrid System connected with Tie-line.

Figure 5. Block Diagram of Diesel Engine Generator Transfer Function model.

6 S. C. ET AL.
Figure 6. Wind Turbine First Order Transfer Function model.
Real-time data from GAMESA power wind turbine model
has been considered in this study. The power output from
the wind turbine has been formulated as the sixth order
polynomial by curve-fitting technique [10] as given in
Equation (2). For small signal stability of the system, the
rate of change of wind power output given in Equation (3)
has been considered for assessing the stability of the pro-
posed systems. Figure 6 represents the block diagram of
the first-order transfer function model of WTG.
2.3. Battery Energy Storage System(BESS)
Due to the high inertia of rotating mass of conventional
frequency regulating device like DEG, it is not suitable
for frequency regulation for dynamic load variations [8].
In view of this, there is a need to introduce a fast and
dynamic frequency regulating device which is BESS
considering sudden load variations. The first-order
transfer function model of BESS is given in Figure 7.
The BESS can be either in charging mode or dischar-
ging mode based on the system frequency as shown in
Table 1.
2.4. Solar Power
Practical solar irradiance data has been obtained from
NREL for a specific duration which has been presented
in section 5. The first-order transfer function model [8]
of Solar PV power is given in Figure 8.
Figure 7. BESS First Order Transfer Function model.
Table 1. Battery charging status based on system frequency.
ΔF BESS status
Positive Charging
Negative Discharging
Figure 8. Solar Power First Order Transfer Function model.
3. Mathematical Problem Formulation
Frequency Deviation of microgrid and multi-microgrid
system and tie-line power flow variation (only in case of
multi-microgrid) are considered as reference for optimal
tuning of PID controller gains. ITAE criteria has been
considered as Fitness function which is the performance
index in fine-tuning the PID controllers gains due to its
advantages of smaller overshoots/undershoots and oscilla-
tions compared to other performance indices such as
integral-squared error (ISE) which gives minimum over-
shoot but more settling time, integral absolute error (IAE)
which produces slower response than ISE in LFC control-
ler design, integral time-weighted-squared error (ITSE)
wherein for sudden change in input, produces larger con-
troller output as stated in reference [10]. The fitness func-
tion ITAE is defined in Equation (5) and the boundaries
for gains of the PID controller are defined in Equation (6).
Fitness ¼ Minimize
n fITAEg
Tsim
o
¼ Min ò 0 t:ðjΔF1 j þ jΔF2 j þ jΔPtieline jÞ:dt
(5)
Subjected to PID gain limits.
9
Kpmin  Kp  Kpmax =
Kimin  Ki  Kimax (6)
;
Kdmin  Kd  Kdmax
Area Control Error (ACE) is given as input for the
PID controller, which is defined as the difference
between the error signal of tie-line flow ΔPtieline and
Bias times the change in respective microgrid system
frequency. For stand-alone microgrid system, the ACE
is simply Bias times the change in frequency as there is
no tie-line. The ACE for multi-microgrid system is
defined in Equation (7).
ACEi ¼ Bi ΔFi þ ΔPtie;ij (7)
It is noteworthy to highlight that for controlling the
power system dynamic, these load frequency control-
lers are designed in off-line mode for planning studies
considering various scenarios before placing them into
online action. Accordingly, GWO controller can be
used for tuning the PID gains during off-line mode
before incorporating in original system operation [10].
4. Grey Wolf Optimization
A novel swarm Intelligence technique, Grey wolf opti-
mization is attempted in this paper for LFC. This
algorithm has been developed from the behavior of
Grey wolves by Seyedali Mirjalili and et al. [30]. Grey
wolves are treated to be the best predators in finding
 SMART SCIENCE 7

Table 2. PID gains obtained for various controllers.

PID-1 gains PID-2 gains
Sl.No Controller Kp Ki Kd Kp Ki Kd
1. Conventional PID 4.3718 1.6559 4.0041 4.4381 1.6995 4.2908
2. PSO-PID 4.9047 0.9176 0.9826 4.9484 0.0034 4.8686
3. TLBO-PID 4.9827 0.9186 2.6694 4.9889 0.0076 4.3625
4. GWO-PID 4.9970 0.9861 4.9485 4.9997 0.0097 4.9992

non-presence of these wolves’ leads to internal fighting

and problems among the pack.

4.1. Mathematical Model of GWO

In this section, the social hierarchy of grey wolves,
tracking, encircling and attaching prey are mathemati-
cally modeled.

4.1.1. Social Hierarchy

In this optimization technique, the first best solution is
considered as alpha(α), beta(β) as the second best and delta
(δ) wolves as the third best and the rest of the solutions of
the pack are considered as omega(ω) wolves. The best
wolves in the pack are being followed by the omega
wolves(ω).

4.1.2. Encircling
Figure 9. Grey wolves hierarchy. The grey wolves encircle the prey during the hunting
process. Encircling the prey is mathematical model as
follows
the prey. An interesting behavior of Grey wolves is D X
 ¼ jC:  P ðt Þ  X
 ðt Þj (8)
their very strict dominant hierarchy of the pack as
presented in Figure 9 is.  ðt þ 1Þ ¼ X D
 P ðt Þ  A: 
X (9)
Alpha is the most powerful wolf in the pack and it can
be either a male or a female. Decisions regarding hunting, where Xp ðt Þ denotes current position of prey and t
migration, sleeping place, feeding are taken by the alpha denotes current iteration.
wolf. An interesting aspect is that the alpha wolf is that it The following equations are used to compute the
must be best in managing the pack but not essentially the values of A and C.
strongest member of the pack which means that organi-  ¼ 2a:r1  a
A (10)
zation and discipline are more important than strength.
Beta wolves are the next level in the hierarchy.  ¼ 2:r2
C (11)
While making decisions, beta wolves assist alpha
wolves. When alpha in the pack is ill or dead, then r1 , r2 are the random vectors in [0, 1]. As number of
they lead the pack. They are treated as discipliner for iterations progresses, the value of a decreases linearly
the pack as well as an advisor to alpha wolves. Delta is from 2 to 0 and is defined as
the third category in the hierarchy. They dominate  
iter
omega but they should report to alpha and beta wolves. a ¼ 2  2 (12)
itermax
Scouts, caretakers, and hunters belong to delta cate-
gory. Omega is the last in the ranking. Being ranked Where iter and itermax represents current iteration
last, they are scapegoat and allowed to eat at last. and maximum number of iterations, respectively. The
The importance of these wolves is that due to the  is [-2a 2a].
boundary of A
8 S. C. ET AL.

4.1.3. Hunting  is randomly generated through the algorithm, the

of C
The ability of the Grey wolves is to recognizing the emphasis is more towards exploration from starting to
prey location and encircling the prey. With the equa- final iteration which avoids local optimum.
tion given in the Encircling phase, the grey wolves
relocate their position to anywhere in the search
4.2. GWO Controller Algorithm for LFC Problem
space. As mentioned above, for survival of the pack
and for hunting strategy, social hierarchy plays a key As mentioned above, in this paper GWO controller has
role. The mathematical model of hunting strategy of been implemented for optimal tuning of PID gains for
the Grey wolves has been described in [25,26,30]. The LFC of stand-alone microgrid and multi-microgrid
formulas for updating the position are as follows. systems. The algorithm implementation steps are
   enumerated as follows.
 ðt þ 1Þ ¼ X 1 þ X 2 þ X 3
X (13)
3
(1) Population size (Psize Þ, itermax, number of con-
1, X
The values of X  2 and X
 3 are calculated as trol variable(n), (ie:; Kp ; KI and KD for each
X  1: D
α  A
1 ¼ X α (14) controller), boundaries of PID controller gains
(KPmin ; KPmax ; KImin ; KImax ; KDmin and KDmax ) to be
where D  1 :X
 α ¼ jC  α  Xj

initialized.
X  2 :D
β  A
2 ¼ X β (15) (2) Generate the population (Pop) randomly within
their limits.
where D  2 :X
 β ¼ jC  β  Xj

(3) Run the Simulink program and evaluate the
3 ¼ X
X  3 :D
δ  A δ (16) fitness (ITAE) values for all the populations
using Equation 5.
where D  3 :X
 δ ¼ jC  δ  Xj

(4) Sort the population according to fitness. Assign
alpha(α), beta(β), delta(δ) & omega(ω) wolves
4.1.4. Attacking Prey(exploitation) based on fitness values.
As stated above, once the prey stops moving, the grey (5) Update the positions of α, β and δ grey wolves
wolves encircle and hunt the prey. The hunting pro- based on fitness value as follows
cess gets finished by attacking the prey. The mathe-
matical model of approaching the prey is modeled by For i ¼ 1 to Psize
decreasing the value of a, which in turn decrease the if ðfitnessðpopðiÞÞ < fitnessðαÞ
fluctuating range of A in the random value interval of
α ¼ popðiÞ
[−2 2]. The exploration of prey is emphasized when
the value of A > 1 or A<-1 and exploitation is empha- end
sized when −1 < A < 1. The algorithms exploration
and exploitation behavior based on the value of A has if ðfitnessðpopðiÞÞ > fitnessðαÞ and fitnessðpopðiÞÞ
been presented by running the program five times in < fitnessðβÞ
the reference [26]. Though the agents update their β ¼ popðiÞ
position for attacking the prey, based on the location
 D
of α, β and δ wolves and the values of parameters A,  end
and a, the algorithm prone to stagnate at local opti- if ðfitnessðpopðiÞÞ > fitnessðαÞ and fitnessðpopðiÞÞ
mum. Hence, there is a need for more parameters for > fitnessðβÞ and fitnessðpopðiÞÞ < fitnessðδÞ
exploration.
δ ¼ popðiÞ
4.1.5. Search for Prey (Exploration) end
In addition to the parameters defined above, there is
end
another parameter favoring exploration of the algorithm
 The values of C
is C.  varies randomly in [0 2] in contrast to (6) Update the position of omega (ω) grey wolves
a, which decreases linearly from 2 to 0. The contribution of using Equations (14–16).
prey in defining the next position of wolves is decided by (7) Update the control variables using Equation 13
 When the value of parameter C
the value of C.  > 1, the and check whether they are violating their lim-
wolves are attracted more towards the prey. Since the value its. If violated, keep within their limits.

Figure 10. Flowchart of GWO Controller algorithm.
(8) Go to step-3 and repeat the above steps until
convergence criteria are satisfied.
4.3. Flowchart for LFC Problem Using GWO
Controller
The flowchart for implementation of GWO control-
ler for optimal tuning of PID gains is shown in
Figure 10.
5. Simulation Results and Analysis
The desirable properties of any control system are
quick response and stability. It is worth mentioning
that higher relative stability and better time-domain
specifications can be obtained with smaller perfor-
mance indices. The acceptability of controller has
been established after making a critical investigation
SMART SCIENCE 9
of system dynamics i.e., transient response specifica-
tions, peak overshoot/undershoot and settling time.
For identification of robustness of a closed loop control
system, the dynamic behavior of microgrid to be eval-
uated with diverse loading conditions and parametric
variations. Accordingly, the dynamic stability of the
system has been investigated with Step load perturba-
tion, with the incorporation of the intermittent nature
of RES and also with parametric uncertainty of the
system.
The proposed GWO controller performance is
scrutinized in finding the optimal gains of PID
Controller on a stand-alone microgrid and multi-
microgrid systems. The simulation work has been
carried out using MATLAB/Simulink software on
Intel Core i3 processor, 4GB RAM, for studying the
performance of the proposed controller. Due to the
stochastic nature of GWO algorithm, the input para-
meters presented in Appendix 2 are selected with an
exhaustive number of trails. The boundaries for the
PID gains are opted between 0 and 5 after making
several trial and errors. The PID gains for various
controllers obtained from simulation studies are pre-
sented in Table 2.
To assess the dominance of GWO controller, the
simulations are carried out on (a) stand-alone microgrid
system and (b) Multi-microgrid system, consisting of
Diesel Generator, BESS, Solar power generation, wind
power generation, and load, respectively. The superiority
of the proposed controller in improving the dynamic
response under various scenarios, a detail comparative
result with various controllers in literature has been per-
formed. The various parameters considered in the system
are presented in Appendix 1.
5.1. Scenario – 1: Single Microgrid – Step Load
Disturbance
In this scenario, a single microgrid with only Diesel
Generator and BESS system has been considered. A
single step load deviation is applied to the system as
presented in Figure 11(a). Figure 11(b) represents the
frequency response of the system. From Table 4, it is
noticeable that the proposed GWO controller is hav-
ing minimum overshoot of 0.047 Hz and minimum
settling time of 21secs (51–30 = 21) in comparison
with other controllers. It is conspicuous from Table 3
that the ITAE values obtained with GWO controller
are minimum compared to other controllers. From
Table 4, it is evident that the peak overshoot/
10 S. C. ET AL.

Figure 11. (a) Step Load fluctuation (b) Single MG frequency response (GWO, PSO, TLBO, and PID).

Table 3. ITAE values obtained for different scenarios using microgrid system consisting of DEG and BESS. The
various controllers. dynamic response of the system obtained for various
Conventional PID PSO-PID TLBO-PID GWO-PID controllers has been presented in Figure 12(b). It is
Scenario-1 0.01898 0.01467 0.01416 0.01394 apparent from Figure 12(b) and Table 5 that magnitude
Scenario-2 0.04954 0.03924 0.03802 0.03750 of oscillations, peak overshoot/undershoot and settling
Scenario-3 0.62851 0.53994 0.51307 0.49770 time has been reduced considerably using GWO con-
Scenario-4 0.14361 0.12929 0.12889 0.12649 troller as against other controllers. It is clear from Table
Scenario-5 0.80508 0.57100 0.55893 0.54620
5 that the peak overshoot/undershoot by GWO control-
Scenario-6 0.15065 0.13056 0.13033 0.12690
Scenario-7 0.83459 0.57857 0.56507 0.55401
ler is 12.43%, 8.82% and 10.40% smaller as against PSO,
TLBO and conventional PID controllers, respectively.

undershoot by GWO controller is 9.62%, 6.00% and
14.55% smaller as against PSO, TLBO, and conven-
tional PID controllers, respectively.
5.2. Scenario – 2: Single Microgrid – Multi-step
Load Disturbance without RES
In this scenario, multi-step load deviation as shown in
Figure 12(a) has been applied to the stand-alone
5.3. Scenario – 3: Single Microgrid – Multi-step
Load Disturbance with RES
In this scenario, isolated microgrid with step load
perturbations, wind power generation, solar power
generation, and BESS has been considered. The plot
for multiple perturbations has been shown in Figure
13. To get better understandability of frequency
response of various controllers, initially PID, TLBO
and PSO controllers are compared. The better

Table 4. Settling time and peak overshoot/undershoot of various controllers for Scenario-1.
Controller GWO-PID PSO-PID TLBO-PID Conventional PID
Disturbance occurred at t = 30secs Settling time in secs 21 26 25 32
Peak Overshoot/undershoot in Hz 0.047 0.052 0.050 0.055
 SMART SCIENCE 11

Figure 12. (a) Multi-Step Load fluctuation (b) Stand-alone MG frequency response (GWO, PSO, TLBO, and PID).

Figure 13. Multiple signal disturbances.

Table 5. Settling time and peak overshoot/undershoot of various controllers for Scenario-2.
Controller GWO-PID PSO-PID TLBO-PID Conventional PID
Disturbance occurred at t = 30secs Settling time in secs 20 24.5 23.5 25
Peak Overshoot/undershoot in Hz 0.0155 0.0177 0.0170 0.0173

controller out of the above three is found to be TLBO
and the same is used for comparison with the pro-
posed GWO controller as shown in Figure 14(a,b). It
is manifested from Figure 14(b) that GWO controller
is having lower peak overshoot/undershoot and set-
tling time as against TLBO controller.
5.4. Scenario – 4: Multi-microgrids – Multi-step
Load Disturbance without RES
In this scenario, multi-microgrids connected with
tie-line have been considered. A step load distur-
bance in microgrid-1 has been initiated. The
12 S. C. ET AL.
(a)
(b)
Figure 14. (a) Stand-alone MG frequency response (PSO, TLBO, and PID) (b) Stand-alone MG frequency response comparison of
TLBO and GWO.
dynamic frequency responses of both the microgrids
and tie-line flow have been shown in Figure 15(a,b,
c). The supremacy of GWO controller as against
PSO, TLBO and PID controllers is evident from
Figure 15 and Table 6 which has less peak overshoot
and small settling time.
5.5. Scenario – 5: Multi-microgrids – Multi-step
Load Disturbance with RES
In this scenario, multi-microgrids system having
wind power generation in microgrid-1 and solar
power generation in microgrid-2 is shown in Figure
4. The step load deviation considered to arise in
microgrid-1. Due to the fluctuating nature of Solar
and wind power generation, continuous oscillations
can be observed in the frequency response of both
the microgrids frequency response including tie-line
flow. To verify the superiority of proposed controller,
initially PID, PSO and TLBO controllers are com-
pared and the obtained response is presented in
Figure 16(i)(a),Figure 16(ii)(a) and Figure 16(iii)(a).
The comparative analysis proves that TLBO is super-
ior over other controllers. Accordingly, TLBO con-
troller is considered to ascertain the ascendancy of
GWO controller as shown in Figure 16(i)(b),Figure
16(ii)(b) and Figure 16(iii)(b).
5.6. Scenario – 6: Multi-microgrids – multi-step
load disturbance without RES including parametric
variation
As stated above, to verify the robustness of any controller,
frequency response due to parametric uncertainty has to
 SMART SCIENCE 13

(a)

(b)

(c)
Figure 15. (a) MG-1 frequency response (GWO,PSO,TLBO and PID) (b) MG-2 frequency response (GWO,PSO,TLBO and PID) (c) Tie-line
flow in P.U between MG-1 and MG-2 (GWO,PSO,TLBO and PID) for multi-step load disturbance.

Table 6. Settling time and peak overshoot/undershoot of various controllers for Scenario-4.
Disturbance considered at t = 10secs
ΔF1 ΔF2 ΔPtieline
Peak Overshoot/ Peak Overshoot/ Peak Overshoot/
Controllers Settling time undershoot in Hz Settling time undershoot in Hz Settling time Undershoot in Hz
GWO 20 0.07553 21 0.0496 24 0.02753
PSO 23 0.08082 24 0.05427 25.5 0.03063
TLBO 21 0.07974 22 0.05385 26 0.02959
PID 28 0.08202 28 0.05844 27 0.03117

be gauged. According, in this scenario, dynamic variation in parameters is presented in Table 7. It is clear
responses of multi-microgrid system with load perturba- from Table 8 and Figure 17 that the proposed GWO
tion and parametric variation have been appraised. The controller is producing minimum oscillations, smaller
14 S. C. ET AL.

(i)

(a)

(b)

(ii)

(a)

(b)
Figure 16. (i).(a) MG-1 frequency response (PSO, TLBO, and PID) (b) MG-1 frequency response comparison of TLBO and GWO, for
multi-step load disturbance including RES. (ii).(a) MG-2 frequency response (PSO, TLBO, and PID) (b) MG-2 frequency response
comparison of TLBO and GWO, for multi-step load disturbance including RES. (iii).(a) Tie-line flow in P.U between MG-1 and MG-2
(PSO, TLBO, and PID) (b) Tie-line flow in P.U between MG-1 and MG-2 comparison of TLBO and GWO, for multi-step load disturbance
including RES.
 SMART SCIENCE 15

(iii)

(a)

(b)
Figure 16. (Continued).

Table 7. Parameter variations.

Sl.No Parameter
1 R
2 D
3 H
settling time and less peak overshoot than other
controllers.
5.7. Scenario – 7: Multi-microgrids – multi-step
load disturbance with RES including parametric
variation
In this scenario, multi-microgrids have been consid-
ered with load perturbations, RES integration, and
parametric variations. The frequency response
% change (ΔF1 ; ΔF2 Þ and ΔPtie of different controllers have
+5% been presented in Figure 18. For better understand-
−25% ability of superiority of proposed controller, first PID,
+30%
PSO, and TLBO controllers are compared as shown in
Figure 18(i)(a), Figure 18(ii)(a) and Figure 18(iii)(a). It
can be analyzed that TLBO controller gives better per-
formance out of the above three controllers.
Considering it, TLBO and GWO controllers are com-
pared and presented in Figure 18(i)(b), Figure 18(ii)(b)
and Figure 18(iii)(b). It is intelligible from the figure
that the GWO controller produces lesser oscillations
and smaller settling time than TLBO controller.
It is evident from the above scenarios that GWO
controller is better in performance such as minimum
oscillations, smaller settling time, lesser peak

Table 8. Settling time and peak overshoot/undershoot of various controllers for Scenario-6.
Disturbance considered at t = 30secs
ΔF1 ΔF2 ΔPtieline
Peak Overshoot/ Peak Overshoot/ Peak Overshoot/
Controllers Settling time Undershoot in Hz Settling time Undershoot in Hz Settling time Undershoot in Hz
GWO 24 0.03553 25 0.02524 24 0.01459
PSO 28 0.03773 28 0.02750 28.5 0.01581
TLBO 27 0.03727 26 0.02739 28 0.01537
PID unstable 0.03942 unstable 0.03097 unstable 0.01617
16 S. C. ET AL.
(a)
(b)
(c)
Figure 17. (a) MG-1 frequency response (GWO,PSO,TLBO and PID) (b) MG-2 frequency response (GWO,PSO,TLBO and PID) (c) Tie-line
flow in P.U between MG-1 and MG-2 (GWO,PSO,TLBO and PID) for multi-step load disturbance and parametric variations.
overshoot/undershoot and minimum ITAE as pre-
sented in Table 3 in comparison with other control-
lers available in the literate. The convergence plot
obtained for scenario-V with the proposed controller
and other prior-art controllers is presented in
Figure 19. It is clear from Figure 19 that GWO
takes less number of iterations in comparison to
other algorithms.
6. Conclusion
In this paper, load frequency control of stand-alone
microgrid and multi-microgrids connected with tie-
line by regulating the PID controller gains embedded
in the individual microgrid system has been
addressed. Grey wolf optimization (GWO) algorithm
has been exercised for generating optimal gains of
PID controller for dynamic stability of the system
under various disturbances such as step load pertur-
bations, sporadic nature of RES integration (wind
power and solar power) and parametric uncertainty
of system. The efficacy and robustness of the pro-
posed GWO controller for stabilizing the system
frequency and tie line flow under various perturba-
tions have been confirmed from the simulation
results. Simulation results obtained using proposed
controller has been compared with conventional PID,
PSO-PID and TLBO-PID controllers to corroborate
 SMART SCIENCE 17

(i)

(a)

(b)
(ii)

(a)

(b)
Figure 18. (i).(a) MG-1 frequency response (PSO, TLBO, and PID) (b) MG-1 frequency response comparison of TLBO and GWO, for
multi-step load disturbance including RES and parametric variations. (ii).(a) MG-2 frequency response (PSO, TLBO, and PID) (b) MG-2
frequency response comparison of TLBO and GWO, for multi-step load disturbance including RES and parametric variations. (iii).(a)
Tie-line flow in P.U between MG-1 and MG-2 (PSO, TLBO, and PID) (b) Tie-line flow in P.U between MG-1 and MG-2 comparison of
TLBO and GWO, for multi-step load disturbance including RES and parametric variations.
18 S. C. ET AL.

(iii)

(a)

(b)
Figure 18. (Continued).

Figure 19. Convergence characteristics of GWO, TLBO, and PSO.

the potential benefits of proposed controller in terms ORCID

of settling time, peak overshoot/undershoot and the
Srinivasarathnam c http://orcid.org/0000-0003-3975-1699
obtained results are validated.

References
Disclosure statement
[1] Dreidy M, Mokhlis H, Mekhilef S. Inertia response and
No potential conflict of interest was reported by the frequency control techniques for renewable energy
authors. sources: A review. Renew Sust Energ Rev. 2017;69:144–
155.

[2] Shankar G, Mukherjee V. Load frequency control of an
autonomous hybrid power system by quasi-opposi-
tional harmony search algorithm. Int J Electr Power
Energy Syst. 2016;78:715–734.
[3] IEEE Guide for Design. Operation, integration of dis-
tributed resource island systems with electric power
systems. IEEE Standard. 2011;1547(4):1–54.
[4]. C. Srinivasarathnam, Chandrasekhar Yammani &
Sydulu Maheswarapu (2019) Multi-Objective Jaya
Algorithm for Optimal Scheduling of Dgs in
Distribution System Sectionalized into Multi-
Microgrids. Smart Science 7 (1):59-78. doi: 10.1080/
23080477.2018.1540381.
[5] Chowdhury AH, Asaduz-Zaman M. Load frequency
control of multi-microgrid using energy storage sys-
tem. 8th International Conference on Electrical and
Computer Engineering; Dhaka, 2014. p. 548–551.
[6] Arefifar SA, Ordonez M, Mohamed YAI. Energy man-
agement in multi-microgrid systems—development
and assessment. IEEE Trans Power Syst. 2017
March;32(2):910–922.
[7] Mallada E, Zhao C, Low S. Optimal Load-Side Control
for Frequency Regulation in Smart Grids. IEEE Trans
Autom Control. 2017 Dec;62(12):6294–6309.
[8] Annamraju A, Nandiraju S. Robust frequency control in
an autonomous microgrid: A two-stage adaptive fuzzy
approach. Electr Pow Compo Sys. 2018;46(1):83–94.
[9] Hajiakbari M, Golshan H, Esmail M. determining
optimal virtual inertia and frequency control para-
meters to preserve the frequency stability in islanded
microgrids with high penetration of renewables. Electr
Power Syst Res. 2018;154:13–22.
[10] El-Fergany AA, El-Hameed MA. Efficient frequency
controllers for autonomous two-area hybrid microgrid
system using social-spider optimiser. IET Gener
Transm Distrib. 2017;11(3):637–648.
[11] Bevrani H, Feizi MR, Ataee S. Robust frequency con-
trol in an islanded microgrid: H∞ and µ-Synthesis
Approaches. IEEE Trans Smart Grid. 2016 March;7
(2):706–717. .
[12] Shi J, Yue D, Huang C, et al. Adaptive distributed
secondary control of microgrids via single-network
adaptive dynamic programming method. Int Trans
Electr Energ Syst. 2018;28:e2549.
[13] Xu Y, Li C, Wang Z, et al. Load frequency control of a
novel renewable energy integrated micro-grid contain-
ing pumped hydropower energy storage. IEEE Access.
2018;6:29067–29077.
[14] MiarNaeimi F, Azizyan G, Rashki M. Multi-level cross
entropy optimizer (mceo): an evolutionary optimiza-
tion algorithm for engineering problems. Eng Comput.
2018;34(4):719–739.
[15] Dhanalakshmi R, Palaniswami S. Load frequency con-
trol of wind diesel hydro hybrid power system using
conventional PI controller. Eur J Sci Res. 2011;60
(4):630–641.
SMART SCIENCE 19
[16] Vidyanandan KV, Senroy N. Frequency regulation in a
wind–diesel powered microgrid using flywheels and fuel
cells. IET Gener Transm Distrib. 2016;10(3):780–788.
[17] Ray PK, Mohanty SR, Kishor N. Proportional–integral
controller based small-signal analysis of hybrid distrib-
uted generation systems. Energy Convers Manage.
2011;52(4):1943–1954.
[18] Mallesham G, Mishra S, Member S, et al. Ziegler–
nichols based controller parameters tuning for load
frequency control in a microgrid. Int. Conf. on
Energy, Automation, and Signal; 2011 Dec 28–30,
Bhubaneswar, Odisha, p. 1–8.
[19] Senjyu T, Nakaji T, Uezato K, et al. A hybrid power
system using alternative energy facilities in isolated
island. IEEE Trans Energy Convers. 2005 June;20
(2):406–414.
[20] Goya T, Omine E, Kinjyo Y, et al. Frequency control in
isolated island by using parallel operated battery sys-
tems applying H∞ control theory based on droop
characteristics. IET Renew Power Gener. 2011;5
(2):160–166.
[21] Jeya Veronica A, Senthil Kumar N. Internal model based
load frequency controller design for hybrid microgrid
system. Energy Procedia. 2017;117:1032–1039.
[22] Lal DK, Barisal AK. Load Frequency Control of AC
Microgrid Interconnected Thermal Power System. IOP
Conf Ser. 2017;225:012090.
[23] Das DC, Roy AK, Sinha N. GA based frequency con-
troller for solar thermal–diesel–wind hybrid energy
generation/energy storage system. Int J Electr Power
Energy Syst. 2012;43(1):262–279.
[24] Kumar RH, Ushakumari S. Biogeography-based
Tuning of PID controllers for load frequency control
in microgrid. Int. Conf. on Circuit, Power and
Computing Technologies [ICCPCT]; 2014 March 20–
21 ; Nagercoil, 2014;. p. 797–802
[25] Gupta E, Saxena A Siew Chong Tan (Reviewing
Editor). Grey wolf optimizer based regulator design
for automatic generation control of interconnected
power system. Cogent Eng. 2016:3(1). doi:10.1080/
23311916.2016.1151612.
[26] Faris H, Aljarah I, Al-Betar MA, et al. Grey wolf
optimizer: a review of recent variants and applications.
Neural Comput Appl. 2018;30(2):413–435.
[27] Pandey SK, Mohanty SR, Kishor N, et al. Frequency
regulation in hybrid power systems using particle
swarm optimization and linear matrix inequalities
based robust controller design. Int J Electr Power
Energy Syst. 2014;63:887–900.
[28] Pahasa J, Ngamroo I. Coordinated control of wind
turbine blade pitch angle and PHEVs Using MPCs
for load frequency control of microgrid. IEEE Syst J.
2016 March;10(1):97–105.
[29] Anil Kumar A, Srikanth NV. Teaching-learning opti-
mization based adaptive fuzzy logic controller for
Frequency control in an autonomous microgrid. Int J
Renewable Energy Research. 2017;7(4): 1842-1849.
20 S. C. ET AL.

Appendix 1.

Wind Turbine parameters : KWTG ¼ 1, TWTG ¼ 1:5

Solar PV system parameters : KPV ¼ 0:0075, TPV ¼ 0:03
BESS : KBES ¼ 1, TBES ¼ 0:1
Valve Actuator : T1 ¼ 0:025, T2 ¼ 2, T3 ¼ 3
Diesel Engine : KE ¼ 1, TE ¼ 3
Speed Regulation Constant : R1 ¼ 5 P:U:HzMW , R2 ¼ 5 P:U:HzMW
Synchronizing power coefficient : T12 ¼ 0:225Π
Rotor Swing-1 : KP1 ¼ 60, TP1 ¼ 18
Rotor Swing-2 : KP2 ¼ 60, TP2 ¼ 18

Appendix 2.

Algorithm PSO TLBO GWO

Common Parameters Psize 50 50 50
itermax 300 300 300
Algorithm specific parameters C1 ¼ 2:0 NIL  C
a, A,
C2 ¼ 2:0
Wmin ¼ 0:1
Wmax ¼ 0:9